Skip to main content
Back to challenges
Challenge

Orchestrate GenAI Medical Device Compliance Audits with AutoGen

Regulatory submission of generative AI-enabled medical software demands stringent safety and verification compliance. Build a multi-agent regulatory reviewer system using AutoGen to inspect LLM software artifact documentation and output automated FDA SaMD risk evaluation matrices with zero false safety passes.

Data ScienceHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Multi-agent conversational audit workflow developed with Microsoft AutoGen for regulatory medical device compliance.

Delivery guide

How work is evaluated

Evaluation

Evaluates regulatory compliance flag precision and completeness in software documentation analysis.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

no_false_pass_check

Ensures incomplete submissions are marked NON_COMPLIANT.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

audit_precision

Precision of identified missing safety controls. • target: 0.95 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Construct multi-agent conversation flows using Microsoft AutoGen `AssistantAgent` and `UserProxyAgent`.

  • Implement regulatory compliance checking algorithms aligned with FDA Software as a Medical Device (SaMD) guidance.

  • Generate comprehensive risk mitigation matrices from technical AI software documentation.

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of 30 synthetic FDA SaMD submission draft summaries containing intentional compliance gaps and correct safety documentation.

Challenge setup

Evaluates regulatory compliance flag precision and completeness in software documentation analysis.

Challenge input

JSON containing system_description, hazard_analysis, and monitoring_plan

AutoGen

Microsoft framework for multi-agent conversational systems.

Evaluated output

JSON object with compliance_status, missing_controls array, and risk_tier

Evaluation checks
  • Ensures incomplete submissions are marked NON_COMPLIANT.
  • Precision of identified missing safety controls. • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: RegAI-Bench
  • Audit Precision target: 0.95
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • AutoGen · Required
  • Aim · Optional
  • Optuna · Optional

Evaluation contract

  • no_false_pass_check · Weight 1
  • audit_precision · Weight 1

Metrics

  • audit_precision · Target 0.95

Benchmarks

  • RegAI-Bench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing system_description, hazard_analysis, and monitoring_plan

    Output: JSON object with compliance_status, missing_controls array, and risk_tier

    Execution harness: Python

Evaluation tools

  • AutoGen · agent_framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Orchestrate GenAI Medical Device Compliance Audits with AutoGen